In an RKE cluster (without a cloud provider configured) is there a way to have Rancher automatically removed worker nodes from the cluster after it's been unavailable/unreachable for X amount of time?
12/20/2022, 9:29 PM
nope. Without an external cloud controller to verify that the instance has actually been deleted, it is generally considered unsafe to assume that a node that is NotReady is gone for good.
12/20/2022, 9:35 PM
Ok thanks. I'm using the cluster-autoscaler in Rancher to scale up/down worker nodes in an EC2 autoscale group. My problem is that when nodes are scaled down and killed in the ASG they're left hanging around in Rancher until I manually delete them. I know this is basically working as intended but I was hoping for a way around it without configuring a cloud provider, it'll be tricky in our environment (IAM permissions are locked down quite a bit).
12/20/2022, 9:37 PM
yeah, I think you’re intended to use the AWS cloud controller to handle that